Acute neuropathological consequences of short-term mechanical ventilation in wild-type and Alzheimer’s disease mice

Mechanical ventilation is considered a life-saving critical care intervention that is implemented in nearly 800,000 patients annually in the USA [1]. It is now recognized that up to 40% of these patients acquire new and significant cognitive impairment that often resembles the phenotype of Alzheimer’s disease (AD), a neurodegenerative condition characterized by cerebral amyloid β (Aβ) peptide accumulation, neuroinflammation, and blood-brain barrier dysfunction [2, 3]. It is also known that the risk for cognitive decline after critical illness is particularly high in individuals with pre-existing cognitive impairment, of which AD is the most common type [4, 5].

It has been hypothesized that mechanical ventilation may contribute to long-term cognitive impairment by promoting cerebral accumulation of the Aβ peptide, systemic and neurologic inflammation, and blood-brain barrier dysfunction although the exact mechanisms remain unknown [6‐9]. To our knowledge, no prior studies have examined the effects of mechanical ventilation in models of AD, which is relevant since most patients who undergo mechanical ventilation are elderly, and thus at higher risk for carrying increased cerebral Aβ pathology [1, 8, 10‐12].

Accordingly, in this study, we sought to test the following hypotheses: (i) short-term injurious mechanical ventilation increases cerebral Aβ accumulation in the double transgenic APP_SWE/PSEN1_∆E9 (ADtg) mouse models of AD, (ii) short-term mechanical ventilation increases systemic and neurologic inflammation in wild-type and ADtg mice, and (iii) short-term mechanical ventilation increases blood-brain barrier permeability of small or large-size molecules in wild-type and ADtg mice.

An overview of the experimental timeline is shown in Fig. 1a. ADtg mice (n = 12; mean age 4.9 ± 0.8 months and weight 26.2 ± 2.9 g) and wild-type mice (n = 8; mean age 5.3 ± 0.8 months and weight 26.3 ± 3.7 g) underwent mechanical ventilation with a tidal volume of 15 ml/kg for 4 h, while control groups of non-mechanically ventilated ADtg mice (n = 9; mean age 4.9 ± 1.8 months and weight 27.8 ± 3.2 g) and WT mice (n = 12; mean age 5.7 ± 1.1 months and weight 27.5 ± 1.2 g) were not mechanically ventilated. There were no significant differences in age or weight between any of the groups (p > 0.05). There was a decrease in mean oxygen saturation during the 4 h of mechanical ventilation in both ADtg and wild-type mice. In addition, the mean oxygen saturation was significantly lower in the wild-type mice compared to the ADtg mice (Fig. 1b). Polymorphonuclear (PMN) cell percent was measured in the bronchoalveolar fluid to assess the extent of pulmonary inflammation due to mechanical ventilation and was significantly increased in the mechanically ventilated wild-type and ADtg mice (Fig. 1c). We then quantified soluble Aβ_1–40 and Aβ_1–42 peptides, which are closely associated with the pathogenesis of AD [3]. There was a significant increase in cerebral concentrations of soluble Aβ_1–40 in age- and sex-matched ADtg mice subjected to mechanical ventilation compared to those who did not undergo mechanical ventilation (Fig. 1e), and there was a direct correlation between soluble Aβ_1–40 and percent PMN cells in the bronchoalveolar fluid (Fig. 1f). There was a non-statistically significant increase in soluble Aβ_1–42 of ADtg mice subjected to mechanical ventilation compared to those who did not undergo mechanical ventilation (Fig. 1d).

Figure 2 illustrates notable changes in cognition-relevant cytokines following mechanical ventilation in both mice strains (see Additional file 1: Figures S2–S7 for all other findings). In comparison to the non-mechanically ventilated mice, mechanically ventilated mice demonstrated significant increases in cerebral IL-6 and TNF-α in both mice strains (Fig. 2a, e). There was a significant increase in IL-5 in mechanically ventilated wild-type mice and a significant increase in IL-10 in mechanically ventilated ADtg mice (Fig. 2m, o). There was a trend towards increased plasma IL-6 following mechanical ventilation in both mice strains while plasma IL-1β increased significantly in the wild-type mice only (Fig. 2b, j). In contrast, there was no significant change in plasma TNF-α following mechanical ventilation and no significant correlation with cerebral TNF-α (Fig. 2f, g). There were direct and significant correlations between (i) PMN percent in the bronchoalveolar fluid and cerebral IL-6, TNF-α, IL-5, and IL-10 concentrations (Fig. 2d, h, n, p) and (ii) plasma and cerebral concentrations of IL-6 and IL-1β (Fig. 2c, k). There were significant interactions between genotype and mechanical ventilation for cerebral IL-5, IL-10, and plasma IL-1β (indicated by p_i in Fig. 2m, o, and j). Other significant interactions can be found in Additional file 2.

Assessment of potential relationships between cerebral Aβ concentrations and cognition-relevant cytokines in ADtg mice revealed direct and significant correlations between (i) cerebral TNF-α and soluble Aβ_1–40 (Fig. 3a), (ii) cerebral IL-10 and soluble Aβ_1–40 (Fig. 3b), (iii) cerebral IL-5 and soluble Aβ_1–40 (Fig. 3c), and (iv) cerebral IL-1β and soluble Aβ_1–40 (Fig. 3d). There was no significant association between cerebral IL-10 and soluble Aβ_1–42 and cerebral IL-1β and soluble Aβ_1–42 (Fig. 3e, f).

Alterations in blood-brain barrier function are implicated in AD and were, thus, assessed by quantifying permeability of small (Texas Red-dextran 3 kD) and large (FITC-dextran 2000 kDa) molecules as evaluated by histology. Although there was a large variability in the sample data for FITC-dextran and Texas Red-dextran (Fig. 4a, b), a significant decrease in the hippocampal accumulation of Texas Red-dextran 3 kD and FITC-dextran 2000 kD were noted in mechanically ventilated ADtg mice and a non-significant increase in hippocampal accumulation of Texas Red-dextran 3 kD and FITC-dextran 2000 kD in the wild-type mice compared to the respective controls who did not undergo mechanical ventilation (Fig. 4a, b). There was a significant interaction between genotype and mechanical ventilation on the measured outcome of Texas Red-dextran 3 kD permeability (Fig. 4b). There was a significant positive correlation between percent of polymorphonuclear cells in the bronchoalveolar fluid and hippocampal accumulation FITC-dextran 2000 kD (Fig. 4d) and a significant negative correlation between cerebral soluble Aβ_1–42 accumulation and fluorescence area of FITC-dextran 2000 kD (Fig. 4e). Representative microscopy with fluorescence staining is shown in Fig. 4c and f. Figure 4c is from one of the three animals who had markedly increased FITC-dextran 2000 kD accumulation following mechanical ventilation while Fig. 4f shows increased cerebral Aβ aggregates (white) in the non-mechanically ventilated ADtg mice associated with hippocampal GFAP+ reactive astrocytes (red) and localized FITC staining (green) in and near blood vessels. The mechanically ventilated mice exhibited reduced hippocampal permeability and 6E10-Aβ deposits near and inside blood vessels (Fig. 4f).

In this study, we found strong correlations between acute pulmonary inflammation after short-term mechanical ventilation and cerebral accumulation of soluble Aβ_1–40, cerebral inflammation, and blood-brain barrier permeability. These data suggest that mechanical ventilation acutely promotes AD neuropathology by (i) increasing cerebral accumulation of the Aβ peptide and AD-associated neuroinflammation, notably TNF-α and IL-6 in ADtg mice with pre-existing AD pathology, and (ii) increasing AD-associated neuroinflammation, notably TNF-α and IL-6, and possibly blood-brain barrier permeability in wild-type mice without pre-existing AD pathology. These findings provide mechanistic insights into the well-documented clinical observations that associate mechanical ventilation with long-term cognitive decline and AD in particular [2, 17]. The non-significant trend towards increased blood-brain barrier permeability in the wild-type mice also suggests the presence of a common pathophysiology that contributes to both delirium, which is independently associated with accelerated cognitive decline, and AD [3, 8‐10, 18, 19]. Further, the finding that soluble Aβ_1–40, the most abundant Aβ isoform, increases with mechanical ventilation is noteworthy because prior studies have shown that soluble Aβ species significantly contribute to AD pathogenesis and even more so than insoluble Aβ aggregates [20‐22].

Although causation is not yet proven through direct experimentation, the results of this investigation are consistent with prior studies that report a role for cerebral TNF-α in promoting cerebral Aβ accumulation [23, 24]; however, this effect is apparently independent of changes in plasma TNF-α, which did not change significantly following mechanical ventilation. This suggests that acute cerebral Aβ accumulation may be independent of systemic TNF-α concentrations and that mechanical ventilation or pulmonary inflammation may directly contribute to neuroinflammation, perhaps via local cerebral signaling mechanisms or by inducing alterations in Aβ clearance from the brain [25].

Our approach to assessing blood-brain barrier dysfunction involved in vivo administration of high (FITC-dextran 2000 kD) and low (Texas Red-dextran 3 kD) molecular weight tracers 30 min prior to sacrifice, followed by tissue extraction and histological analysis. Notably, the molecular weight of Aβ_1–42 is approximately 4.4 kD and is thus closely approximated by the size of the Texas Red-dextran 3 kD tracer influx that was used in this study [26]. This approach allowed us to reveal the complex relationships between mechanical ventilation, cerebral Aβ accumulation, and blood-brain barrier permeability. Our results show a trend towards increased hippocampal blood-brain barrier permeability in the wild-type animals, which suggests that short-term mechanical ventilation may impair the function of neuroanatomical structures responsible for cognition in otherwise healthy brains. Since the blood-brain barrier plays a central role in maintaining normal cognition via clearance of Aβ and maintaining cerebral small vessel integrity, these findings suggest that mechanical ventilation may initiate neuropathology that leads to long-term cognitive impairment [8‐10, 27‐30]. Paradoxically, the ADtg mice showed significantly decreased blood-brain barrier permeability despite increased cerebral Aβ_1–40 concentration. These unexpected results may suggest that Aβ exerts a protective effect that limits the influx of toxins from the vascular space into the brain parenchyma across the blood-brain barrier. Future studies are needed to assess whether leakage of tracer is more restricted in regions where Aβ are aggregated and to clarify whether this is related to altered expression of Aβ receptors or transporters within the blood-brain barrier.

This study has important limitations that should be considered. Although these data indicate statistical significance in three principal components of cognitive impairment—namely, accumulation of cerebral soluble Aβ_1–40, increase in key cognition-relevant inflammatory biomarkers, and alteration of blood-brain barrier permeability—it is possible that future studies with larger sample sizes could reveal additional significant effects of mechanical ventilation on other Aβ isoforms that are less abundant than soluble Aβ_1–40. Conversely, it is also possible that the trend towards increased blood-brain brain permeability in the wild-type mice reverses with larger sample sizes, notwithstanding the biological plausibility of increased blood-brain barrier permeability in systemic and systemic inflammatory conditions [7]. Additionally, as there were significant interactions between genotype and mechanical ventilation for cerebral IL-5, IL-10, and Texas Red-dextran 3 kD permeability, future studies should be designed to identify genotype-specific variables that may have influenced these outcomes.

Another potential limitation is the use of high dose of mechanical ventilation, i.e., 15 cc/kg tidal volumes, which may not reflect current clinical practice where lower tidal volume ventilation is generally considered the standard of care. However, these animals were also healthier and did not have any concurrent systemic infections or exposure to high doses of sedatives that could have increased the overall medical acuity. Subsequent research is needed to determine whether lower tidal volume ventilation decreases the burden of neurological injury or whether superimposed effects of systemic diseases such as sepsis alter the extent of neurological injury. Although the young mice used in this experiment do not necessarily reflect the demographic of older intensive care patients who undergo mechanical ventilation. The use of these animals with less baseline AD-pathology allowed us to clearly demonstrate the accelerated trajectory of AD-relevant neuropathology. Future studies are needed to study the effects of mechanical ventilation in older mice.

To simulate the real-life application of mechanical ventilation, the animals who underwent mechanical ventilation received sedation, while the non-mechanically ventilated animals did not receive sedation. The provision of sedation to the non-mechanically ventilated control group could alter respiratory drive and lead to additional confounding variables, such as hypoxemia or hypercapnia, which could independently alter cerebrovascular hemodynamics and the measured outcomes. We, thus, opted to avoid this effect of sedation in our control animals. The issue of sedation remains challenging and one that will need to be clarified in future investigations that strive to identify the independent effects of sedatives on the measured outcomes. Finally, the results of this study must be interpreted with caution as findings from pre-clinical studies are not invariably confirmed in clinical trials, and although significant efforts were made to minimize introduction of bias, including the use of blinded data analysts and concealed experimental group allocation, it may not be possible to fully eliminate the introduction of bias.

In conclusion, our data indicate that short-term mechanical ventilation, a common critical care intervention, promotes neuropathology that is a characteristic of AD by increasing cerebral soluble Aβ_1–40, promoting neuroinflammation, and altering blood-brain barrier permeability. Future studies are needed to further clarify specific molecular mechanisms that underlie these findings that will facilitate the development of neuroprotective mechanical ventilation strategies that could mitigate the risk of cognitive decline after critical illness.